STIR.AcquisitionSensitivityModel.forward etc can crash python

[KrisThielemans]

stir.BinNormalisation.apply etc can throw exceptions (e.g. when not being able to write to file in the example below). They are currently not caught, see e.g. here, but there are many other places),
leading to python just exiting:

>>> m=sirf.STIR.AcquisitionData('SIRFmultfactors_withatn_3.hs')
>>> asm.set_up(p)
>>> xx=asm.unnormalise(p)

WARNING: write_data: error after writing to stream.


ERROR: ProjDataFromStream::set_viewgram: viewgram (view=0, segment=-11, timing_pos=-5) corrupted due to problems with writing or the scale factor (out of disk space?)
terminate called after throwing an instance of 'std::runtime_error'
  what():  ProjDataFromStream::set_viewgram: viewgram (view=0, segment=-11, timing_pos=-5) corrupted due to problems with writing or the scale factor (out of disk space?)
Aborted
$ bash>

Looks like we need to TRY...CATCH stuff around these calls.

[evgueni-ovtchinnikov]

@KrisThielemans can I have your Python script and your data please so I could reproduce the crash?

[KrisThielemans]

sorry, as it crashed, I don't have the history anymore, but essentially

p = pet.AcquisitionData('some-read-only.hs')
n = pet.AcquisitionData('norm.hs')
asm= pet.AcquisitionSensitivityModel(n)
n.set_up(p)
asm.unnormalise(p)

[evgueni-ovtchinnikov]

n.set_up(p) is obviously a mistake - should be asm.set_up(p)

error messages are missing in asm.[un]normalise methods of AcquisitionSensitivityModel, which cannot be applied to a read-only object (there are alternative forward/invert methods that return the [un]normalised AcquisitionData object)

will do a PR now